1

Running Head: SOUND CHANGE IN HONG KONG CANTONESE

Phonetic Variations and Sound Changes in Hong Kong Cantonese:

Diachronic Review, Synchronic Study, and Implications for Speech Sound Assessments

Carol K. S. To

The University of Hong Kong, Hong Kong SAR

Sharynne McLeod

Charles Sturt University, Australia

Pamela S. P. Cheung

Child Assessment Service, Hong Kong SAR

Address for correspondence: Carol To, Division of Speech and Hearing Sciences, Faculty of

Education, 7/F Meng Wah Complex, The University of Hong Kong, Pokfulam Road.

Email: tokitsum@hku.hk

Tel:852-39171591

Fax:852-25590060

To appeared in Clinical Linguistics & Phonetics (2015)

2

Abstract

The aim of this paper was to describe phonetic variations and sound changes in Hong Kong

Cantonese (HKC) in order to provide speech-language pathologists with information about

acceptable variants of standard pronunciations for speech sound assessments. Study 1

examined the pattern of variations and changes based on past diachronic research and

historical written records. Nine phonetic variations were found. Five in syllable-initial and

syllabic contexts: (1)[n-]→[l-], (2)[ŋ-]→Ø-, (3)Ø-→[ŋ-], (4)[kwɔ-]→[kɔ-], (5)syllabic

[ŋ̍]→[m̩]; and four in syllable-final contexts: (6)[-ŋ]→[-n], (7)[-n]→[-ŋ], (8)[-k]→[-t],

(9)[-t]→[-k]. Historical records demonstrated the pattern of variation and changes in HKC

across time. In study 2, a large-scale synchronic study of speakers of differing ages was

undertaken to determine acceptable phonetic variations of HKC for speech sound

assessments. In the synchronic study single-words were elicited from 138 children (10;8-12;4)

and 112 adults (18-45 years) who spoke Cantonese and lived in Hong Kong. Synchronic

evidence demonstrated five acceptable variants in syllable-initial and syllabic contexts: (1)

[n-]→[l-], (2) [ŋ-]→Ø-, (3) Ø-→[ŋ-], (4) [kwɔ-]→[kɔ-], and (5) syllabic [ŋ̍]→[m̩] and four

incomplete sound changes in syllable-final contexts: (6) [-ŋ]→[-n], (7) [-n]→[-ŋ], (8)

[-k]→[-t], and (9) [-t]→[-k]. The incomplete sound changes may still be accepted as variants

in speech sound assessments unless related speech problems are indicated.

3

Phonetic Variations and Sound Change

Sound change occurs in all living languages. For example, Old English was derived from

the Germanic language subgroup which in turn stemmed from the Indo-European language

family. At present, different varieties of English are spoken in Britain, the United States and

other English-speaking speech communities all over the world. They vary in their vowel and

consonant systems, for example, post-vocalic [ɹ] is extensively used in the United States but

does not occur in most of England. Within each English-speaking speech community, sounds

continue to change over time. For example, Labov (1994) reported that post-vocalic [ɹ] was

introduced into New York City as a prestige variable in the years following World War II. He

conducted a survey in three department stores of different social status in New York City in

1962 and found that only some upper middle class speakers used post-vocalic [ɹ] in casual

speech while people of lower middle and lower classes used it in more formal speech styles

only. Fowler (1986, cited in Labov 1994) replicated Labov’s earlier study and noted there was

an increase in the percentage of speakers of all social classes using post-vocalic [ɹ].

Phonetic variations among speakers in a speech community are the key element causing

sound changes. The initial basis for sound changes begins from subtle phonetic variability in

speech among speakers in a community. These variations can originate from various phonetic

factors such as coarticulation, listeners’ perception, and speakers’ intention in varying their

speech for different communicative functions, or social factors such as gender. The ‘new’

variants may extend across linguistic and social contexts. Once variants are established as

commonplace within a speech community and become part of the sound system, sound

changes are said to have been completed (Hock, 1991). In other words, synchronic variation

is a pre-condition of sound change (Ohala, 1989). Previous studies have revealed consistency

between synchronic variations among speakers in a speech community and sound changes

4

manifested in cross-language tendencies in segmental inventories (Ohala, 1989).

Phonetic Variations and Sound Change in Hong Kong Cantonese (HKC)

Cantonese belongs to the Yue dialect group and is spoken in southeast China as well as

by many Chinese communities all over the world. Cantonese is the language spoken as the

“usual language” by 89.5% people in Hong Kong (Census and Statistics Department,

HKSAR Government, 2012), a city with a population of 7 million and a land area of 1,104

square kilometers. Different Cantonese or Yue varieties are spoken in places around Hong

Kong. In terms of phonology, these varieties differ from each other in the phonemic and tone

system to different extents (Zhan & Y.-S Cheung, 1987). The syllabic structure of Hong Kong

Cantonese (HKC) is (C)V(C) with segmental elements of an onset and a rime and a

supra-segmental element of tone. There are 19 initial consonants /p-, ph-, t-, th-, k-, kh-, kw-,

kwh-, ts-, tsh-, f-, s-, h-, w-, j-, l-, m-, n-, ŋ-/ and six final consonants /-p, -t, -k, -m, -n, -ŋ/.

The 8 vowels are /i, y, ɛ, œ, а, ɔ, u, ɐ/, and 11 diphthongs are /аi, ɐi, аu, ɐu, ei, ɵy, ɔi, ui, iu, ou,

ɛu/. In addition, Cantonese has nine tones: Tone 1 “High level”; Tone 2 “Low-mid to high

rising”; Tone 3 “Mid level”; Tone 4 “Mid-low falling”; Tone 5 “Mid-low rising”; Tone 6

“Mid-low level”, Tone 7 “High stopped”; Tone 8 “Mid stopped” and Tone 9 “Mid-Low

stopped” (To, Cheung, & McLeod, 2013; Zee, 1999a).

In the past century, the speech sounds of Hong Kong Cantonese (HKC) have been

changing. In a highly densely populated city like Hong Kong, it is not surprising that sound

changes can disseminate rapidly through people’s close interaction as well as the mass media.

Children learn these variants from their parents, peers, television, and the internet. Within

Hong Kong there are different opinions and attitudes regarding phonetic variations and sound

changes. While the general public is not very aware of the current sound changes in HKC,

some educators advocate the use of the standard (dictionary) pronunciations because they

5

consider these changes or variations as a result of “laziness” and insist that “correct

pronunciations” need to be conserved (Ho, 1995). At the same time, some linguists regard

sound changes and variations as natural phenomena and are interested in documenting

non-standard pronunciations within Hong Kong society.

Importance of Documenting Acceptable Variants for Speech-Language Pathologists

Throughout the world when speech-language pathologists (SLPs) assess children’s

speech to determine whether or not they have speech sound disorders, they rely on a standard

set of pronunciations of the ambient language, and, consequently, for each language a set of

pronunciations are defined. For example, Smit et al. (1990) provided a list of acceptable

phonetic variants to the standard pronunciation of General American English in the Appendix

to their large normative study of children’s speech sound acquisition. An example was the use

of glottal stop [ʔ] for word-final /t/, as in cat [khæʔ] in place of /khæt/.

SLPs working with children in Hong Kong, as well as those working with

Cantonese-speaking children in other countries also need to define standard acceptable

pronunciations. Traditional speech sound assessments of Hong Kong Cantonese (HKC) are

based on standard pronunciations as recorded in dictionaries (e.g., So, 1993). SLPs working

with Cantonese-speaking children have clinical concerns about differentiating speech errors

from phonological variants resulting from sound change. Anecdotally, most SLPs in Hong

Kong make allowances regarding acceptable variants according to their personal judgment. It

is important for SLPs to apply consistent and present-day definitions of speech sound

production regarding whether or not realizations can be accepted as phonemic variants or

should be recorded as speech sound errors.

The present study aimed to provide an evidence-base to define standard pronunciations

that reflects present-day HKC pronunciation. Such up-to-date information would enable SLPs

6

to accurately assess a child’s speech sound production ability and allocate therapy resources

to those who have true speech sound disorders rather than those who produce acceptable

variants.

STUDY 1: DIACHRONIC REVIEW

The aim of this first study was to document historical sound change in HKC using

previously published work. Three sources of evidence were available to describe recent

phonetic variations in HKC. The first source comprised descriptive records (e.g., Ball, 1907;

IPA, 1949; Li et al., 1995; S.-L. Wong, 1941). The second source comprised reviews (e.g.,

Bauer & Benedict, 1997; H.-N. Cheung, 1972). The third source comprised fieldwork studies

on speakers of HKC most of whom were born in Hong Kong (e.g., Bourgerie, 1990; Chen,

1999; Lin, 1995; S.-Y. Wong, 2005; Yeung, 1980). Additionally, Zee (1999b) provided

empirical palatograms and linguagrams data for one 21-year-old speaker of HKC. The

findings from these three sources present the progress of variations in HKC in the past few

decades. Finally, since Cantonese belongs to the Yue dialect, similar phonetic variations in

other Yue varieties near Hong Kong as recorded in A Survey of Dialects in the Pearl River

Delta (Zhan & Y.-S. Cheung, 1987) were examined. From a wider perspective, other Chinese

dialects as recorded in a survey undertaken by the Beijing University (1989) were also

reviewed. Both surveys were undertaken in the 1980’s and the pronunciations recorded in the

syllabaries could be considered conservative as the informants selected were from the

middle-aged generation to older people in their eighties. If similar changes also occur in other

Yue speech communities or even in other Chinese dialects, this may suggest that certain

inherent phonemic characteristics of these cognate varieties and dialects, which are prone to

changes, are reacting in a similar way when conditions that trigger variants appear.

Consideration of these three sources of evidence revealed nine recent phonetic variations in

7

HKC consonants (see Table 1) which will be reviewed in the following sections. ---------------------------------

Insert Table 1 about here ---------------------------------

Phonetic variation 1: Replacement of [n-] with [l-].

The phonemes /n/ and /l/ differ by the presence of nasality and an involvement of the

tongue blade for /n/. According to the phonetic description by Zee (1999a, p.60), [n] is

“apico-laminal denti-alveolar” while [l] is an “apical denti-alveolar or apical alveolar”.

Historical records and reviews. The oldest preserved record of Middle Chinese

phonology is the rhyme book Guangyun, which was published in 1008 A.D. for standard

rhyming in poetry. Both Cantonese and Middle Chinese in Guangyun share a common

ancestry (Li et al., 1995) and there is a close relation between the two sound systems. The

rhyme book Guangyun distinguished /n-/ and /l-/ in syllable-initial position.

Nine hundred years later there was still a clear distinction between /n-/ and /l-/ in

standard Cantonese (Ball, 1907). Replacing [n-] with [l-] was first mentioned in the early

1940’s by S.-L. Wong (1941) who regarded the new (non-standard) realization as a

pronunciation error. At this time the International Phonetic Alphabet (IPA, 1949) recognized a

clear distinction between /n-/ and /l-/ words in HKC and did not mention replacement of [n-]

with [l-]. In the 1970’s, H.-N. Cheung (1972) noted that replacement of [n-] with [l-] was

very common in HKC.

Fieldwork studies of speakers of HKC. Since the 1970’s there have been four studies of

speakers of HKC that have described the replacement of [n-] by [l-]. Yeung (1980) reported

that for 50 participants (aged 12-57 years; i.e., born between the 1920’s and 1960’s), the

replacement of [n-] by [l-] occurred more in males and in younger speakers. In fact, of the

youngest group of speakers (born in the 1960s), 100% of their productions had replaced [n-]

8

by [l-]. The speakers’ home dialects did not affect the variance.

Bourgerie (1990) also found a significant age gradation in her 49 participants.

Replacement of [n-] by [l-] was produced 28.2% of the time by participants aged 46-87 (born

between the 1900’s and 1940’s); however, this figure tripled to 85.4% in the youngest age

group of 6-18 years (born in the 1970’s and 1980’s). Only four out of his 49 informants had

pure [n-] productions and they were all in their eighties (born in the 1900’s). The other two

80+-year-old informants had a mixture of [l-] and [n-]. Different from Yeung’s (1980)

findings, Bourgerie’s female speakers used [l-] more than the male speakers, but the gender

difference was not significant.

The replacement of [n-] with [l-] occurred more often in more recent studies. For

example, Chen (1999) reported that her fourteen 12-year-old female students (i.e., born in the

1980’s) produced 92.9% of [n-] words as [l-]. Based on the data from the palatograms and

linguagrams by Zee (1999b), the 21-year-old male participant (born in the 1970’s) only used

[l-] for [n-].

Summary. To summarize, replacing [n-] by the non-standard [l-] in HKC started to grow

around the 1900’s and [l-] has been adopted by more than 80% of the population born in and

after the 1970’s. The change correlates with age but gender difference has been inconsistent.

This variation has also been found in other Yue varieties of nearby regions as well as in other

Chinese dialect groups. While in some dialect varieties the change is complete, some are

having the change in progress when both [l-] and [n-] realizations are present in the speech.

Phonemically, the merging of [n-] into [l-] reduces one syllable-initial sound contrast, so that

spoken syllables only begin with [l-].

Phonetic variations 2 and 3: Replacement of [ŋ-] with zero initial Ø- and

replacement of zero initial Ø- with [ŋ-].

9

In standard Cantonese, [ŋ-] and zero initial Ø- are in complementary distribution. Zero

initial Ø- appears in words with mid to high tones, namely, T1, T2, T3, T7, and T8, while [ŋ-]

appears in words with low to mid tones, namely, T4, T5, T6, and T9 (Wong, S.-L., 1941).

Historical records and reviews. Similar to [n-] and [l-] words, Ø- and [ŋ-] words of

present-day standard Cantonese have maintained the same initials as in Middle Chinese a

thousand years ago (Wang, 1985; Li et al., 1995). One hundred years ago Ø- and [ŋ-] were

clearly distinguished in standard Cantonese (Ball, 1907) except that a few colloquial mid to

high tone syllables adopted [ŋ-] instead of Ø-, for example, [ŋɐp7] to talk widely, [ŋɐi1] to

importune. Low tone syllables using Ø- were not found.

In the 1940’s S.-L. Wong (1941) explained that [ŋ-] could be added before Ø- syllables

beginning with [а, ɐ, o, ɔ] in standard Cantonese. Subsequently, H.-N. Cheung (1972) noted

that most HKC speakers overused [ŋ-] for Ø- except for a few particles, for example, particle

for exclamation [ai]. Thus, from the historical records, it appears that Cantonese speakers

favored [ŋ-] for both Ø- and [ŋ-] initial syllables at least until the early 1970’s.

Fieldwork studies of speakers of HKC. Since the 1970’s there have been five studies of

speakers of HKC that have described the replacement of [ŋ-] with zero initial Ø- (phonetic

variation 2) and the replacement of zero initial Ø- with [ŋ-] (phonetic variation 3).

Yeung (1980) studied 50 participants aged between 12 and 57 years. For phonetic

variation 2, the older participants (aged 20-57, born between the 1920’s and 1950’s) replaced

[ŋ-] with Ø- less than 15% of the time whereas younger participants (aged 12-16, born in the

1960’s) replaced [ŋ-] with Ø- 56.8% of the time. Male speakers used Ø- more. However,

reverse trends were found in phonetic variation 3 and this change appeared to diminish with

time: the older participants aged 20 and over replaced Ø- with [ŋ-] between 26.4% and 28.8%

of the time while the younger participants born in the 1960’s had this change for 11.8% of the

10

time only. It is apparent that participants born in the 1960’s used Ø- more than [ŋ-] no matter

the standard form was initial [ŋ-] or Ø-.

Bourgerie (1990) examined phonetic variation 2 in 49 people aged 6-87 years. The older

participants (31-87 years, born before 1960) only replaced [ŋ-] with Ø- 9.5% to 9.9% of the

time; whereas, the younger participants (6-30 years, born after 1960) replaced [ŋ-] with Ø-

36.7% to 66.2% of the time. This phonetic variation was significantly correlated with age

and gender (with female having more Ø- realizations), but not with education or place of

origin. A similar trend was noted by Lin (1995) who studied phonetic variation 2 in 50 people

aged 12-63 years. The older participants (born before the mid-1960’s) replaced [ŋ-] with Ø-

0% to 3.7% of the time; whereas, the younger participants (born after the 1960’s) replaced

[ŋ-] with Ø- 58.9% to 62.1% of the time. Lin (1995) also studied phonetic variation 3 in the

same 50 people and found the reverse trend that the older participants (born before the

mid-1960’s) replaced Ø- with [ŋ-] 67.0% to 86.4% of the time; whereas, the younger

participants (born after the 1960’s) had the same replacement reduced to 24.8% and 26.8% of

the time. Only age but not gender, educational level or occupation was significantly

correlated with both phonetic variations. For Chen’s (1999) fourteen 12-year-old female

participants (born in the mid-1980’s), Ø- productions took up 77.4% of the standard [ŋ-]

items in phonetic variation 2. Finally, the palatographic and linguagraphic data of the

21-year-old male showed that he produced 66.7% of [ŋ-] items as Ø- (phonetic variation 2)

and 34.6% of Ø- items as [ŋ-] (phonetic variation 3) (Zee, 1999b).

To summarize, the development of syllable-initial Ø- and [ŋ-] in HKC has changed

over the past hundred years. Consistent findings show that in the first half of the 20th century

speakers favored the production of [ŋ-] for both syllable-initial Ø- and [ŋ-]; however, the

more simple Ø- has grown in popularity among speakers born after 1960. This means that

11

phonetic variation 2 is growing rapidly while phonetic variation 3 is diminishing. Age has

been correlated with these changes but findings regarding the effect of gender are

inconsistent.

Summary. In the past hundred years that speakers favored the production of [ŋ-] for

both initials in the first half of the 20th century but speakers born after the 1960’s adopted the

more simple Ø-. Age has been correlated with the change but findings in gender are

inconsistent. The simplification process of using Ø- for both initials has been completed in

two out of the 25 Yue speech communities though merging towards the more complicated [ŋ-]

was also found in one Yue region as recorded in the Pearl River Delta survey in the 1980’s

(Zhan & Cheung, Y.-S., 1987). When initial [ŋ-] disappears in HKC, the number of initial

consonants will be reduced by one.

Phonetic variation 4: Replacement of [kw-] and [kwh-] with [k-] and [kh-] before [ɔ].

Unlike the phonetic variations discussed before, the change from [kw-] to [k-] and [kwh-]

to [kh-] is conditioned phonetically. That is, labial dissimilation occurs when the two

labialized velars precede the mid back round vowel [ɔ].

Historical records and reviews. There was no record that [kw-] and [kwh-] were

delabialized before [ɔ] in standard Cantonese a hundred years ago (Ball, 1907) or in the

1940’s (S.-L. Wong, 1941; IPA, 1949). The gradual disappearance of the labialization of [kw-]

and [kwh-] before [ɔŋ] and [ɔk] was first mentioned by H.-N. Cheung (1972) in the 1970’s.

Fieldwork studies of speakers of HKC. There have been two studies of speakers of

HKC that have considered phonetic variation 4. Among the 75 participants of Bauer’s (1983)

sociolinguistic study undertaken in 1979-1981, the use of the variation from [kw-] to [kɔ] had

a clear age gradation. In spontaneous speech, the older participants (aged 45-75, born

between 1905 and 1935) produced the phonetic variation [kɔ] 56.6% of the time; the next age

12

group (aged 31-44, born between 1936 and 1949) produced [kɔ] 68.5% of the time; the next

age group (aged 23-30, born between 1950 and 1957) produced [kɔ] 92.6% and the youngest

age group (aged 15-22, born between 1958 and 1965) produced [kɔ] 97.7% of the time.

Slightly over half of the speakers over the age of 30 (born before 1950) had mixed

productions of both the standard and nonstandard forms in their speech and only two speakers,

aged 68 and 73, in the cohort carefully preserved the standard form in spontaneous speech.

Chen (1999) studied fourteen 12-year-old female students (born in the mid-1980’s) and

found that [kwɔ] was replaced by [kɔ]) 67.9% of the time.

Summary. The historical studies examining phonetic variation 4, the dissimilation of [kw]

and [kwh] before [ɔ], demonstrate that this change appeared in the 1900’s and is now

widespread in present-day HKC. The change has merged [kwɔ-] and [kwhɔ-] words into [kɔ-]

and [khɔ-] words. This merge is also common in other Yue varieties.

Phonetic variation 5: Replacement of syllabic [ŋ̍] with syllabic [m̩].

In Cantonese, syllabic [ŋ̍] and [m̩] can exist alone to represent an entire word; for

example, syllabic [m̩] is an adverb meaning not. These two nasal syllabics differ in the place

of articulation with the bilabial place allowing syllabic [m̩] easier to see and articulate than

the syllabic velar [ŋ̍].

Historical records and reviews. In Modern Chinese syllabic [ŋ̍] was suggested to have

been developed from the syllable [ŋ̍u] while syllabic [m̩] was only found in the colloquial

form of dialects (Wang, 1985). In standard Cantonese, the distribution of syllabic [ŋ̍] and

syllabic [m̩] remained unchanged in the first half of the 20th century (Ball, 1907; S.-L.Wong,

1941) and syllabic [m̩] only occurred in the syllable [m̩4] which means not (Wong, S.-L.,

1941).

Fieldwork studies of speakers of HKC. There have been three studies of speakers of

13

HKC that have considered phonetic variation 5. Bauer (1983, 1986) was the first to mention

the replacement of syllabic [ŋ̍] with syllabic [m̩]. He found that there was a clear age

gradation in the use of phonetic variation 5 by his 75 participants. Speakers aged 45-75 (born

between 1905 and 1935) had a mere 0.5% production of the non-standard [m̩] in spontaneous

speech, while those aged 31-44 (born around 1936 and 1949) produced [m̩] 45.7% of the

time, those aged 23-30 (born between 1950 and 1957) produced [m̩] 58.5% of the time, and

those aged 15-22 (born between 1958 and 1965) produced [m̩] 86.6% of the time. A gender

difference was found in the age group of 31-44 years whereby female participants had a much

higher production of syllabic [m̩] in spontaneous speech. Bauer (1986) suggested that the

realization of [m̩] started to appear in the mid-1940’s as a phonetically-conditioned variant

formed by assimilation when the frequently used word five occurred next to a bilabial sound;

for example, [ŋ̍5 mɐn1]→ [m̩5 mɐn1] five dollars, [sɐp9 ŋ̍5] → [sɐp9 m̩5] fifteen, and later

spread to other syllabic [ŋ̍] words through lexical diffusion. A consistent finding was reported

by Chen (1999) in which 92.9% of the productions for syllabic [ŋ̍] was pronounced as the

non-standard [m̩] by the fourteen female participants born in the mid-1980’s.

Summary. In HKC, adopting the simpler syllabic [m̩] for syllabic [ŋ̍] has had its growth

since the 1940’s and is used frequently in present-day speech. Age and gender have found to

be correlated with this phonetic variation. The non-standard form is also found in a number of

Cantonese varieties. The number of syllabic phonemes is reduced from two to one as a result

of phonetic variation 5.

Phonetic variations 6-9: Replacement of [-ŋ] with [-n] and [-n] with [-ŋ], and

replacement of [-k] with [-t] and [-t] with [-k].

There are only six syllable final consonants in Cantonese but phonetic variations have

been reported in four of them. They are the replacement of [-ŋ] with [-n] (phonetic variation

14

6), [-n] with [-ŋ] (phonetic variation 7), [-k] with [-t] (phonetic variation 8), and [-t] with [-k]

(phonetic variation 9). These phonetic variations appeared to be mediated by the preceding

vowel.

Historical records and reviews. In Middle Chinese syllable-finals /-n/, /-ŋ/, /-t/, and /-k/

were distinctive phonemes (Wang, 1985) which are still present in the present-day standard

Cantonese (Li et al., 1995). These four syllable-finals were recorded a hundred years ago

(Ball, 1907) and in the 1940’s (IPA, 1949; Wong, S.-L., 1941).

Fieldwork studies of speakers of HKC. The first study to report syllable-final phonetic

variation was Bauer (1979) who reported the alveolarization (fronting) of syllable-final [-ŋ]

to [-n] (phonetic variation 6) in a young female speaker born in Hong Kong. The female

produced a complete change for syllable-final [aŋ] and [ɐŋ], almost complete change for [ɛŋ],

occasional change for [ɔŋ] and [œŋ], and no change for [ɪŋ] and [ʊŋ]. Yeung (1980) studied

50 people aged 12-57 and only found alveolarization of the syllable-final consonants

(phonetic variations 6 and 8) in those under 27 years of age (born in and after the 1950’s)

with a variant production of 10% at most. Yeung reported phonetic variation 6 in words

having rimes [ɪŋ] and [ɐŋ], and phonetic variation 8 in words having rimes [ɪk] and [ɐk]. No

gender difference was found in the change. Yeung suggested that three out of the four

speakers who produced these phonetic variations could have been affected by their Hakka

and Gan dialects.

Zee (1999b) provided palatographic and linguagraphic data to demonstrate phonetic

variations 6 and 8 in a native 21-year-old male HKC-speaker who realized 33.3% of the [aŋ]

items as [an], 51.7% of the [ɐŋ] items as [ɐn], 53.1% of the [ak] items as [at], and 85.7% of

the [ɐk] items as [ɐt] while other [-ŋ] and [-k] rimes remained unaffected.

Similar findings were reported by Chen (1999) who studied 40 participants aged 13-65

15

(born between the 1930’s and 1980’s). For phonetic variation 6, alveolarization of [-ŋ] was

most frequently found in the vowel contexts [ɐŋ] and [aŋ] and least found in [ʊŋ] and [ɪŋ],

and this varied according to the age of the participants. The variants increased with the

decrease of age and it ranged from 2.6% in the oldest group (aged 40-65) to 20.5% in the

youngest group (aged 13-18). For final [-k] (phonetic variation 8), there was also an age

gradation with the youngest group (aged 13-18) having the most alveolarization (13.5%) and

the oldest group (aged 40-65) having the least (1%). Similar to [ŋ] in phonetic variation 6, the

most common vowel context for phonetic variation 8 was [ɐk] and [ak] and the least were [ɪk]

and [ʊk]. There was little gender difference in the overall variant productions in phonetic

variations 6 and 8.

S.-Y. Wong (2005) examined syllable-final [-ŋ] (phonetic variation 6) in her 40

participants aged between 16-48 (born between the 1950’s and the 1980’s). Similar to the

previous studies phonetic variation 6 occurred least (4.9%) in the older participants (aged

over 41) and most (24.1%) in the younger participants (aged 11-20). It appeared more

frequently in the vowel contexts [ɐŋ] and [aŋ] and least in [ɪŋ] and [ʊŋ]. No gender

difference was found.

Phonetic variations 7 and 9 refer to the process of velarization (backing) of [-n] to [-ŋ]

and [-t] to [-k]. These phonetic variations were only reported by Zee (1999b) in the

palatographic and linguagraphic data of his 21-year-old participant (i.e., born in the 1970’s).

He realized 71.4% of [ɔn] items as [ɔŋ] (phonetic variation 7), 31.3% of [at] items as [ak]

and all [ɔt] items as [ɔk] (phonetic variation 9) while all other [-n] and [-t] rimes remained

unchanged.

Summary. Compared with other phonetic variations in HKC, alveolarization of [-ŋ] to

[-n] and [-k] to [-t] started to grow at a later time of around the 1950’s and the reverse

16

process of velarization of [-n] to [-ŋ] and [-t] to [-k] was first reported in a participant born in

the 1970’s (Zee, 1999b). No gender difference has been mentioned and the changes seem to

occur more frequently in certain vowel contexts. Both alveolarization and velarization have

also found in other Yue varieties and Chinese dialects. However, in HKC, it is interesting to

find that these two contrasting processes occur together.

STUDY 2: SYNCHRONIC STUDY

The diachronic study reported in study 1 revealed nine phonetic variations as emerging

in previous historical records and review studies (see Table 1). This second study aimed to

determine whether these nine emerging phonetic variations noted in the diachronic study

were complete or still emerging in the speech of adult and child speakers in Hong Kong.

Method

Participants

The adult group included 112 participants who came from a variety of professions.

There were 53 males and 59 females aged between 18 and 45 (born between 1960 and 1987)

(see Table 2). The inclusion criteria for the adult participants were that they were born in

Hong Kong in or after 1960, spoke HKC as their mother-tongue, had parents who spoke

HKC without another Chinese dialect or non-Chinese background (as reported by the

participants), and had no previous training in Cantonese phonology to avoid deliberate or

careful production of HKC during the testing. This age range was chosen to ensure that the

participants spoke the present-day variety of HKC. The influx of Chinese immigrants ceased

after the 1950’s; therefore, the adult group represented the Hong Kong-born generations who

were brought up and educated locally, and spoke HKC since birth. The age range spanned

over two decades to capture as many variations as possible encountered within the

community for defining the acceptable variants in present-day Cantonese. The child

17

participant group included 138 Cantonese-speaking students who came from three primary

schools and were aged between 10;8 and 12;4. There were 70 boys and 68 girls (born

between 1992 and 1993). They represented the generation who grew up in the internet era

when multimedia were easily accessible. The parents were requested to fill out a

questionnaire on demographic information including the language use with parents and the

household income. All the parents for the child participants reported that they communicated

with their children in Cantonese. Household income was indicated by three levels according

to the census data (Census and Statistics Department, 2007), namely, < HK$10,000”,

“HK$10,000 – HK$30,000” and “> HK$30,000” and the percentage distribution was 25%,

53%, and 19% respectively.

----------------------- Insert Table 2 here

------------------------

Materials and Procedures

Speech samples were elicited using the beta version of the HKCAT (Cheung, Ng, & To,

2006), a picture naming format with 42 familiar words represented as colour photos bound

into a stimulus book. In the present study, a total of 27 items were selected for the

examination of the phonetic variations in HKC. The selected stimuli were elicited as single

words which were mono-, di-, or trisyllabic.

The elicited stimulus to examine phonetic variation 1 ([n-]→[l-]) was [nam4 tsɐi2] ‘boy’,

phonetic variation 2 ([ŋ-]→ Ø-) was [ŋɐu4] ‘cow’, phonetic variation 3 (Ø-→[ŋ-]) was [ʊk7]

‘house’, phonetic variation 4 ([kwɔ-]→[kɔ-]) was [phɪŋ4 kwɔ2] ‘apple’ and phonetic variation

5 (syllabic [ŋ̍]→[m̩]) was [ŋ̍5] ‘five’. The item [kwhɔ-] was not included as [kwhɔ-] words are

scarce and not easily presented by pictures. As more stimuli with different vowels followed

by [-ŋ] and [-n] were available in the articulation test, phonetic variation 6 ([-ŋ]→[-n]) was

18

examined using the following six words: [phɪŋ4 kwɔ2] ‘apple’, [tshi4 kɐŋ1] ‘spoon’, [thɔŋ2]

‘candy’, [khʊk7 khei4 pɛŋ2] ‘cookie’, [hœŋ1 tsiu1] ‘banana’, and [tshaŋ2 sɪk7] ‘orange color’.

To obtain a complete set of Cantonese [-ŋ] rimes for the [-ŋ] stimuli in which [ʊŋ] was

lacking, the child participants’ production of the non-word [phʊŋ1] was added from the

Nonword Repetition Test with Pseudosyllables (NRTP) (T’sou et al., 2006). In the NRTP,

each participant was asked to imitate the nonword after listening to a pre-recorded model.

The NRTP was administered only to the child participants and not to the adult participants.

Phonetic variation 7 ([-n]→[-ŋ]) were examined with seven words which contained a

complete set of Cantonese [-n] rimes and they were [lai1 lin2] ‘zipper’, [jyn4] ‘round’, [tan2]

‘egg’, [kwhɐn4] ‘dress’, [sɵn3] ‘letter’, [mun4] ‘door’, and [kɔn1] ‘dry’. Phonetic variation 8

([-k]→[-t]) was examined via [tshaŋ2 sɪk7] ‘orange color’, [kœk8] ‘feet’, [ʊk7] ‘house’, and

[lɔk9 jy5] ‘raining’. Phonetic variation 9 ([-t]→[-k]) was examined via [jit9] ‘hot’, [sœ4 wɐt9

thɐi1] ‘slide’, [hɔt8] ‘thirsty’, and [jɐt7] ‘one’. The number of items used to examine the

phonetic variations in the present synchronic study was restricted as only a single test word

was used for each of the first five phonetic variations and four to seven items were used for

the variations related to final consonants. We acknowledge that it is an important limitation of

the present study but the data still provided a preliminary picture about the sound variation

situation in HKC.

The participants were tested individually in a quiet place by 11 qualified SLPs with

training on the standardized administrative procedures of the HKCAT and the NRTP. Each

speaker was requested to say the target words once clearly. The samples were recorded on a

minidisk recorder (either Sony MZ-B100 portable MD recorder or Sharp MD-MT290H(S)

MD recorder) using an external microphone (either ECM-DS70P Sony electret condenser

microphone or RP-VC200 Panasonic stereo microphone) placed about 15 cm away from the

19

mouth of the participant. Each participant’s identity and speech productions were recorded on

separate tracks. The SLPs transcribed the target production on-line using narrow transcription

for phonemes that were produced differently from the targets and marked the transcription on

a separate recording form for each participant.

Analysis

After the testing, two of the 11 SLPs with over 10 years of experience in assessing child

speech and language re-transcribed and scored all the speech samples with narrow

transcription based on the on-line transcription of the recording form. If there was a

discrepancy between the on-line transcription and the recorded speech, either one of the

experienced SLPs made the final decision with reference to the recorded speech. The mean

percentage of the nine non-standard variants of [n-], [ŋ-], Ø-, [kwɔ], syllabic [ŋ], [-ŋ], [-n],

[-k], and [-t] was computed for both the child and the adult groups. In addition, productions

by the two genders within each age group were also examined. Chi-square tests with Yate’s

correction were conducted to investigate whether the differences observed between the

groups were significant. The patterns were compared and discussed with reference to the

previous study in the diachronic study.

Reliability

In order to determine the inter-rater reliability between the two transcribers, samples

from 10 children and 10 adults (8.0% of the speech samples) were randomly selected and

independently scored by both transcribers. The total number of phonemes and tones under

examination were 4,900. Point-to-point agreement between the two transcribers was 98.3%.

Results

Table 3 summarized the occurrence of phonetic variations produced by the adult and

child groups which were also stratified by gender. -----------------------

20

Insert Table 3 here ------------------------

Phonetic variation 1: Replacement of [n-] with [l-].

Over 90% of the participants in the child group (94.2%) and adult (94.6%) realized [n-]

as [l-]. The difference between the two age groups did not achieve statistical significance as

shown in the chi-square tests. Both genders produced more than 90% of the non-standard

form [l-] and no significant gender difference was found in the two age groups. The results of

this variation were consistent with previous studies (Bourgerie, 1990; Chen, 1999; Yeung,

1990; Zee, 1999a) and the change appears to be close to completion.

Phonetic variations 2 and 3: Replacement of [ŋ-] with zero initial Ø- and

replacement of zero initial Ø- with [ŋ-].

For phonetic variation 2 ([ŋ-]→ Ø-), 37.5% of the adults produced Ø- and the figure

significantly raised to 65.0% in the child group. The adult pattern was similar to previously

reported trend that participants born after 1960 replaced [ŋ-] with Ø- from 37% to 66% of the

time (Bourgerie, 1990; Lin, 1995; Yeung, 1980). The younger group in the present study was

also similar to the patterns observed in the young participants in Chen (1999) and Zee (1999a)

who produced more than 60% of the standard [ŋ-] as Ø-. For the phonetic variation 3

(Ø-→[ŋ-]), 31.3% of the adults and only 5.8% of children produced this realization and the

difference was significant. The adult figure was similar to Lin’s cohort born after the 1960’s

(24.8% to 26.4% of the time). The findings suggested that the null realization (Ø-) is

progressively gaining popularity and the tone rules governing the use of [ŋ-] and Ø- no

longer hold. The occurrence of both variations was not associated with gender in the adult

group while in the child group, significantly more boys used Ø- for [ŋ-] than girls.

Phonetic variation 4: Replacement of [kw-] with [k-] before [ɔ].

Half of the adults (50.9%) in the present study showed dissimilation of [kw] to [k] before

21

[ɔ] and significantly fewer children (37.7%) used the variant. No gender difference was

observed in either the child or adult groups. The percentages reported in previous studies

observed in the cohorts born between 1960 and 1980 producing the delabialized non-standard

form were higher than the figures in the present study (97.7% in Bauer (1983); 67.9% in

Chen (1999)).

Phonetic variation 5: Replacement of syllabic [ŋ̍] with syllabic [m̩].

The majority of the adults (95.5%) children (98.6%) produced the non-standard form [m̩]

and no gender difference was observed in either group. The figures were generally consistent

with Bauer (1983) and Chen (1999) who reported that their cohorts born after 1950’s

produced syllabic [m̩] for standard [ŋ̍] more than 85% of the time. This variation appeared to

be close to completion.

Phonetic variations 6-9: Replacement of [-ŋ] with [-n] and [-n] with [-ŋ], and

replacement of [-k] with [-t] and [-t] with [-k].

The phonetic variation 6 ([-ŋ]→[-n]) ranged from 0% to 37.5% in the adult group and

0% to 51.9% in the child group. The changes were dependent on the vowel context as found

in previous studies. The rise in the percentage of the non-standard form of [-n] follows the

vowel order of ʊ = ɪ < ɔ < ɛ < ɐ < a < œ in both the adults and children. For phonetic variation

7 [-n]→[-ŋ], there were only average percentages of 1.3% and 4.0% of [-ŋ] production in the

adult and child groups respectively and the change was much milder when compared to

[-ŋ]→[-n]. Both the adult and the child groups tended to preserve the standard form [-n]. The

vowel context of [ɔ] elicited the highest rate of velarization with 7.1% and 21.9% in the adult

and child groups respectively. There was no gender difference in either group.

Similar to [-ŋ]→[-n], alveolarization of [-k] to [-t] (phonetic variation 8) rarely occurred

before [ɪ] and [ʊ] in all participants. More non-standard productions were observed in the

22

contexts of [ɔ] (20.9% & 14.5%) and [œ] (8.0% & 6.6%) in the adult and child groups

respectively. For the reverse pattern of [-t]→[-k] (phonetic variation 9), [-t] in [ɔt] were more

prone to be velarized to [-k] with 24.1% of the adults and 38.7% of the children producing

the non-standard form.

Overall Discussion

As estimated, based on historical analysis (study 1), the HKC phonetic variations in [n-]

(phonetic variation 1) and [kw-] and [kwh-] before [ɔ] (phonetic variation 4) were first noticed

among speakers born around the 1900’s, syllabic [ŋ̍] (phonetic variation 5) in the mid-1940’s,

[-ŋ] (phonetic variation 6) and [-k] (phonetic variation 8) in the 1950’s, Ø- and [ŋ-] (phonetic

variations 2 and 3) in the 1960’s, and [-n] and [-t] (phonetic variations 7 and 9) probably in

the 1970’s. From the current synchronic study (study 2), the first five sound changes were

frequently used by the locally-born HKC speakers while the last four were emerging and had

increased in use by the young generation in general. These variations have appeared and

progressed quickly within the past century. Three out of the 19 initial Cantonese consonants

(Ø- is counted as an allophone of [ŋ-]), four out of the six final consonants and one out of the

two syllabic nasals have undergone different degrees of changes. What is happening to the

language so that the variations in the sound system can occur quietly but rapidly over the past

century? The changes can be both externally and internally motivated.

External factors

External motivation for phonetic variation and sound change may include contact with other

languages and dialects as well as gender differences.

Language contact.

The population in Hong Kong has expanded dramatically within the past one and a half

centuries from several thousand to seven million. Hong Kong has received continuous

23

influxes of immigrants speaking different Cantonese varieties as well as other Chinese

dialects. The figure shown in the 2006 Hong Kong population by-census that 5.7% of the

total population over 5 years of age used Cantonese as “another language/dialect” reflects the

situation that 1 in 17.5 residents did not speak Cantonese as their mother tongue but had

learned this language for communication (Census and Statistics Department, HKSAR

Government, 2007). Together with residents who speak other Cantonese varieties (they

usually claim that they speak Cantonese as their mother tongue), these non-HKC Cantonese

speakers may produce speech which has retained different variants that sound close to HKC.

Most of the phonetic variations discussed here have already existed in other Cantonese

varieties in regions near Hong Kong (Zhan & Cheung, 1987; Zee, 1999a) as well as in other

Chinese dialects (Beijing University, 1989). Therefore, local Hong Kong born HKC speakers

may also be influenced by non-HKC speakers. With frequent trade and migration, sound

variations in Hong Kong can transmit very rapidly within a short period of time.

Another example of the influence of language contact on language variation occurs with

school children in Hong Kong. Over the past decade the Education Bureau has advocated the

policy of trilingualism (i.e., Cantonese, Putonghua, and English) and biliteracy (i.e., Chinese

and English) in primary and secondary schools. When comparing the performance of the

children and adults in the current study, it is interesting to find that the child group produced a

similar percentage of standard form [n-] in [nam] (phonetic variation 1) as the adult group

and had significantly more occurrences of the standard form [kw] in [kwɔ] (phonetic variation

4) than the adult group. In other words, adults used the variant forms more often than the

children. The trends of the two variations appear to have slowed down than expected. The

distribution of Cantonese initial consonants [n-] and [l-] in the reading pronunciations of the

standard Chinese characters is generally the same as in Putonghua where they are also

24

contrastive phonemes. For example, the word ‘male’ in both Cantonese [nam4] and

Putonghua [nan] have the same [n-] while ‘blue’ in Cantonese [lam4] and Putonghua [lan]

have the same [l-]. By the same token, Cantonese initial consonant [kw-] and [kwh-] have

similar distribution in Putonghua corresponding to [ku-] and [khu-], in which the vowel [u]

also contains a lip-rounding feature similar to [w]. For example, ‘fruit’ is [kwɔ2] in Cantonese

and [kuo] in Putonghua. As Putonghua has been taught more extensively in kindergartens to

secondary schools for more than a decade and internal school assessment on its proficiency is

required, this national language of China may have facilitated school children to make a

distinction between Cantonese [n-] and [l-], as well as to maintain the lip-rounding element [w]

in Cantonese [kwɔ-] and [kwhɔ-] syllables. As a consequence, the rate of change might have

been slightly reduced (Bauer & Benedict, 1997). However, this speculation will require

stronger evidence.

Gender differences.

Previous studies have reported inconsistent results for gender differences of sound

variations in HKC and no gender took the lead. In study 2, there was no significant gender

difference in the results of the adult group for all the phonetic variations under examination.

In the child group, there was a significant gender difference only in [ŋ-]→Ø- and final [-ŋ]

following [ɔ, a, œ] (phonetic variations 2 and 6) such that boys demonstrated the more

nonstandard form Ø- while girls had more variations in [-ŋ]. As both genders in the past few

decades have shared equal educational and job opportunities in Hong Kong, the social

environment should not induce any impact on speech sound acquisition and learning. As a

result, even though a mild significant gender difference was shown in some of the variant

productions, neither gender took the lead.

Internal Factors

25

Internal motivations for the phonetic variations are those that originate within a dialect

due to phonetic factors such as coarticulation, listeners’ perception, lexical obsolescence,

morphological or syntactic structures, or the nature of the writing system. The findings from

the present study mainly shed light on the role of phonetic factors.

Phonetic factors.

By analyzing the patterns of the nine sound variations in HKC, the principle of the

maximum ease of articulation may provide account for some of the more preferential use of

the variant forms which involve less articulatory effort. According to Ladefoged (2006), “the

main way to reduce articulatory effort is by making use of coarticulations between sounds. As

a result of coarticulation, language change” (p. 262). Speakers tend to underarticulate and

conserve articulation effort in producing speech sounds when they are speaking casually. The

context-specific alveolarization process noted in final consonants [-ŋ] and [-k] (phonetic

variations 6 and 8) might be accounted for by the process of coarticulation. The

alveolarization to [-n] and [-t] may be facilitated by a preceding mid-front or central vowel,

that is, [ɛ, œ, ɐ, a]. After the production of these vowels, the tongue moves less when the

front part is raised towards the alveolar region for [-n] and [-t] than when it is backed and

raised to the velar region to produce the standard [-ŋ] and [-k]. On the other hand, the

movement of the tongue to transition from the back vowel [ɔ] or high vowels [ɪ] and [ʊ] to

the high back velar position of [-ŋ] and [-k] is less than to [-n] and [-t], which holds back

alveolarization. Although [ɪ] is a front vowel, it is produced with the tongue nearer to the

central than to a front position and raised just above the half-close position. This place of

articulation is closer to the place when the tongue root is raised for [-ŋ] and [-k] than when

the tip (and blade) of the tongue is raised for the alveolar [-n] and [-t]. This logic also holds

in explaining the velarization of [ɔn] to [ɔŋ] and [ɔt] to [ɔk] (phonetic variations 7 and 9). It

26

is easier for the back of the tongue to be raised slightly from the mid back position of [ɔ] to

produce the non-standard [-ŋ] or [-k] than moving up and forward to produce the standard [-n]

and [-t] at the alveolar position.

The syllabary of Cantonese along with assimilation also provides an explanation for

some of the vowel-specific variations related to final consonants. Phonetic variations 7

([-n]→[-ŋ]) and 9 ([-t]→[-k]) only exist in relation to the [ɔ] vowel. That is, phonetic

variation 7 exists in [ɔn] syllables: [ɔn, ŋɔn, hɔn, kɔn] while phonetic variation 9 exists in [ɔt]

syllables: [kɔt] and [hɔt]. The back position of [ŋ-, h-, k-] and initial Ø- (i.e., glottal stop [Ɂ])

together with the back vowel [ɔ] facilitate progressive assimilation to the final consonant

within a syllable (though this is not legal in Cantonese). Although alveolarization and

velarization are two opposite processes, these two variations occur at the same time in HKC

and are favored by the younger generation who produced significantly more non-standard

forms than the adults not only in [ɛŋ, ɐŋ, aŋ, œŋ] (phonetic variation 6) but also in [ɔn]

(phonetic variation 7) and [ɔt] (phonetic variation 9). These might exist as a complimentary

distribution of [ɪ, ʊ, ɔ] combining with final [-ŋ] (and [-k]) and the other vowels combining

with [-n] (and [-t]) in future development. To investigate this, a post-hoc analysis was

conducted to examine the within speakers’ pattern of the six items ending in [-ŋ]. Of the

speakers who had three or more instances of alveolarization of [-ŋ] to [-n] (i.e., ≥50%), 20%

of the adults and 25% of the children also had velarization of [ɔn] substituted by [ɔŋ] in [kɔn]

(phonetic variation 7) but did not display alveolarization of [-ŋ] in [thɔŋ] (phonetic variation

6). Thus, for these participants the complimentary pattern was observed reflecting the

influence of vowel contexts influencing the phonetic variations.

In terms of articulatory gestures, syllabic [m̩] is more easily produced than syllabic [ŋ̍]

(phonetic variation 5). The non-standard [m̩] only requires keeping the tongue and the lips at

27

rest positions while when producing the standard form [ŋ̍], one needs to raise the dorsal part

of the tongue and drop the lower jaw. Another possible factor that may contribute to the fast

transmission among speakers of the variation is that bilabial closure provides salient visual

cues for the listeners making the phoneme easier to perceive.

For [kwɔ] (phonetic variation 4), labial dissimilation to [kɔ] occurs when speakers retain

the round nuclear vowel [ɔ] and drop the labialized feature of [w]. The same process also

applies to the aspirated counterpart [kwhɔ]. In fact, one of the most distinctive features in

Cantonese phonotactics is the strong tendency in labial dissimilation constraint that

“precludes the combination of labially-articulated initial and final consonants and vowels

within the same syllable” (Bauer & Benedict, 1997, p.281).

Besides production, other researchers also have examined sound changes and phonetic

variations from the perspective of perception and proposed that listeners’ misperception is the

primary source of sound changes. For example, Ohala (1986) regarded dissimilation as a

form of hypercorrection by the listeners. In order to extract essential phonetic information,

listeners may assign a phonetic feature from a sequence of segments to one single segment.

That means, listeners may wrongly assume a single phonetic source for several surface sound

patterns. In this case, we speculate that the round feature which is characterized by the

labialized feature of [kw, kwh] and [ɔ] is interpreted by the listener as from a single segment

[ɔ]. The result of this monosegmental assignment leads to labial dissimilation of [kw] to [k] in

the condition of [ɔ] vowel. Similarly, for phonetic variations 8 ([-k]→[-t]) and 9 ([-t]→[-k])

which involved the alveolar and velar final stops, evidences have shown that even normal

native Cantonese listeners exhibited confusion when identifying final alveolar and velar stops

(Khouw & Ciocca, 2006; Law, Fung, & Bauer, 2001). Such confusion may be the result of

similar acoustic characteristics of these two phonemes (Khouw & Ciocca, 2006). However,

28

the limited scope and data of the present synchronic study cannot provide further evidence to

support the perceptual claim.

A logographic language.

The writing system of Chinese may indirectly reinforce the sound change process in

HKC. Unlike English which is an alphabetic language with letters carrying clear

phonological information, Chinese script is composed of characters and is a logographic

writing system. Unlike alphabetic languages, each Chinese character does not represent

individual sounds that make up the spoken form of that character; therefore, limited cues are

carried for pronunciation. HKC speakers can read Chinese words in Cantonese pronunciation.

However, the component parts of the logographic written language do not provide any

phonetic information to alert HKC speakers that the new realizations are different from the

standard pronunciations (e.g., [n-] & [l-]). In China, there is a Pinyin system (i.e., standard

Romanization for modern Chinese words) for Putonghua and students start to learn Pinyin at

seven years of age when they enter school. In Hong Kong, students do not need to learn the

pronunciation of HKC through a Romanization system. In fact, the Linguistic Society of

Hong Kong has developed a Romanization system for Cantonese called Jyut Ping which has

been widely adopted by linguists in Hong Kong in addition to the IPA system. While

educators have put a lot of efforts to alleviate the “lazy pronunciation” problem and promote

“proper pronunciation” via developing teaching materials and TV programs (e.g., Ho, 2001),

introduction of a Rominization system like Jyut Ping into the school syllabus may directly

assist students’ perception and production of the speech sounds if they can visualize the

pronunciation of Cantonese alphabetically.

Implications for Speech Sound Assessments

Analysis of diachronic records (study 1) and synchronic variations of HKC (study 2)

29

have provided clear directions for identifying acceptable variants in Cantonese speech

production for use during SLP assessments and interventions. Typical language learners

acquire sounds from the environment and are different from individuals with speech sound

disorders who lack an ability to produce the conventional forms with reference to appropriate

models. Therefore, in an unbiased speech assessment tool, these popular variants identified in

the current paper should be taken into account when assessing a speaker’s phonological

skills.

To summarize, the first five synchronic variations (in syllable-initial and syllabic

contexts) can be used as acceptable productions in speech sound assessment of HKC: (1)

[n-]→[l-], (2) [ŋ-]→Ø-, (3) Ø-→[ŋ-], (4) [kwɔ-]→[kɔ-], and (5) syllabic [ŋ̍]→[m̩]. The four

syllable-final sound changes are not yet used by the majority of the population and so can be

accepted as variants but with caution: (6) [-ŋ]→[-n], (7) [-n]→[-ŋ], (8) [-k]→[-t], and (9)

[-t]→[-k]. Some specific vocabulary items relevant to sound changes 6 to 9 have high

adoption by local speakers. For example, one third of the adults and many more of the

children substituted [-ŋ] by [-n] in [kɐŋ] ‘spoon’, [tshaŋ] ‘orange’ and [hœŋ tsiu] ‘banana’.

The children also assimilated [kɔn] ‘dry’ to [kɔŋ] and [hɔt] ‘thirsty’ to [hɔk] at an

accelerating rate. Speakers may have learned these realizations from the ambient language.

When speakers demonstrated the variants of (6) to (9), SLPs should further examine if the

patterns are general to all vowel contexts. If these variants are specific to certain vowel

contexts as illustrated in this study, it is more likely that these non-standard productions are

the results of these ongoing sound changes and not an indicator of speech sound disorders.

In addition to these phonetic variations, linguists have documented mergers of the two

contrastive tones (high rising and low rising tones) in speakers born in the 1990’s (e.g., Fung

& Wong, 2011). For example, the high rising tone Tone 2 (25) was found to be merging with

30

the mid-low rising tone Tone 5 (23) (Bauer, Cheung & Cheung, 2003; Fung & Wong, 2011).

It is therefore expected that the sound system of HKC will continue to change within the

speech community.

SLPs need to be sensitive towards phonetic variants and to differentiate between

productions reflecting sound changes versus clinical speech errors. In clinical assessments

SLPs should not treat variants as “error” tokens if they exist within the regional standard. On

the other hand, for clients who have fronting and backing errors for initial consonants and

even vowels, SLPs should analyze more carefully whether alveolarization or velarization in

final consonants has the same clinical root as the initial consonants.

Limitations and Future studies

While studies 1 and 2 provide the first comprehensive and integrated data on phonetic

variation and sound change in HKC, there are some limitations that need to be considered.

Firstly, the data collected in the synchronic study (study 2) were restricted by the limited

number of test items in HKCAT. This is not enough for a detailed investigation as the

phonetic environment of a syllable may play a part in determining the readiness or resistance

of the changes. A more extensive study with a sufficiently large number of lexical words of

different frequencies of occurrence, different meanings, different syllable structures, and with

different tones on speakers with finer age group differentiation would provide greater details

about the rate and extent of the use of phonetic variations. Additionally, picture labeling in

the synchronic study (study 2) might have elicited words in a style that encouraged formal

speech compared with the spontaneous speech style. It is expected that formal speech may

elicit fewer non-standard forms than spontaneous speech as shown in Bauer (1983) and S.-Y.

Wong (2005). On the other hand, Chen (1999) found that formality of speech style did not

have a significant impact on the productions of her participants. During the synchronic data

collection, it was interesting to find that some adults were unable to produce the standard

31

forms even when taught explicitly afterwards. As the sound changes in HKC can be

considered as an unconscious change, some speakers may not show much production

difference between the conversational or careful reading styles. It would be worthwhile to

have empirical data on the variation of sound changes in casual and formal speech and the

difference may reflect how much public awareness there is for the standard sounds.

Summary

The current paper provides an overview of literature (study 1) and data from 122

participants (study 2) to document phonetic variation and sound change in HKC. A summary

of the sound changes to be considered by SLPs during assessment of Cantonese-speakers is

found in Table 1. Phonetic variations are continuously emerging in HKC and sound change is

an ongoing process. In addition to the 9 variants described above, Tone 2 and Tone 5 have

been documented to be merging in HKC though this pattern was not observed in the current

data set. Future studies using stimuli with minimal pairs and with the help of acoustic

analysis in addition to perceptual judgment may reveal new variation patterns.

References

Ball, J. D. (1907). Cantonese made easy (3rd ed.) . Singapore: Kelly & Walsh.

Bauer, R. S. & Benedict, P. K. (1997). Modern Cantonese phonology. Berlin, Germany:

Mouton de Gruyter.

Bauer, R. S. (1979). Alveolarization in Cantonese: A case of lexical diffusion. Journal of

Chinese Linguistics, 7(1), 132-141.

Bauer, R. S. (1983). Cantonese sound change across subgroups of the Hong Kong speech

community. Journal of Chinese Linguistics, 11(2), 301-354.

Bauer, R. S. (1986). The microhistory of a sound in progress in Hong Kong Cantonese.

32

Journal of Chinese Linguistics, 14(1), 1-42.

Bauer, R. S., Cheung, K.-H., & Cheung. P.-M. (2003). Variation and merger of the rising

tones in Hong Kong Cantonese. Language Variation and Change, 15(2), 211-225.

Beijing University (1989). Hanyu Fangyin zihui. [Chinese dialectal vocabulary] (2nd ed.)

Beijing, China: Wenzi Gaige Chubanshe.

Bourgerie, D. S. (1990). A quantitative study of sociolinguistic variation in Cantonese. Ann

Arbor, MI: UMI.

Census and Statistics Department (2007). Proportion of population aged 5 and over able to

speak selected languages/dialects 1996, 2001 and 2006. In 2006 Population By-census.

Retrieved from:

http://www.bycensus2006.gov.hk/FileManager/EN/Content_981/all4e.xls.

Census and Statistics Department, HKSAR Government. (2012). Population Aged 5 and

Over by Usual Language, 2001, 2006 and 2011 (A107); Population Aged 5 and Over by

Place of Birth, Usual Language and Ability to Speak Other Language/Dialect, 2011

(A123). Retrieved from http://www.census2011.gov.hk/en/main-table/A107.html;

http://www.census2011.gov.hk/en/main-table/A123.html.

Chen, N. L.-L. (1999). The velar coda variation in Hong Kong Cantonese. Unpublished

Masters Thesis. City University of Hong Kong, Hong Kong.

Cheung, H. N. (1972). Xianggang Yueyu yufa de yanjiu [Cantonese as spoken in Hong Kong].

33

Hong Kong: Chinese University of Hong Kong.

Cheung, S. P. P., Ng, K. H. A., & To, C. K. S. (2006). Hong Kong Cantonese Articulation Test.

Hong Kong: Language Information Sciences Research Centre, City University of Hong

Kong.

Fung, R. S. Y. & Wong, C. S. P. (2011). Acoustic analysis of the new rising tone in Hong

Kong Cantonese. Proceedings of the 17th International Congress of Phonetic Sciences,

716-718. Retrieved from

http://www.icphs2011.hk/resources/OnlineProceedings/RegularSession/Fung/Fung.pdf

Ho, W. H. (1995). Yueyin jiaoxue jishi [Records of the career of my Cantonese Chinese

Pronunciation teaching]. Hong Kong: T. T. Ng Chinese Language Research Centre.

Ho, W. H. (2001). Yueyin zi xue tigang [An outline for the self-study of Cantonese Chinese

pronunciation]. Hong Kong: Hong Kong Education Publishing Company.

Hock, H. H. (1991). Principles of historical linguistics (2nd ed.) Berlin, Germany: Mouton de

Gruyter.

IPA (1949). The principles of the International Phonetic Association. London, UK:

International Phonetic Association.

Khouw, E. & Ciocca, V. (2006). An acoustic and perceptual study of final stops produced by

profoundly hearing impaired adolescents. Journal of Speech, Language, and Hearing

Research, 49, 172-185.

34

Labov, W. (1994). Principles of linguistic change. Oxford, UK: Blackwell.

Ladefoged, P. (2006). A course in phonetics. Southbank, Australia: Thomson, Wadsworth.

Law, S. P., Fung, R., & Bauer, R. S. (2001). Perception and production of Cantonese

consonant endings. Asia Pacific Journal of Speech, Language and Hearing, 6, 179-195.

Li, X.-K., Huang, J. J., Shi, Q. S., Mai, Y., & Chen, D. F. (1995). Guangzhou fangyan yanjiu

[The study of the Guangzhou dialect]. Shaoguan, China: Guangdong Renmin

Chubanshe.

Lin, J.-P. (1995). Xianggang Yueyu Ø- ŋ- bianyi yanjiu [Study of variation in Ø- and ŋ- in

Hong Kong Cantonese]. Paper presented at fifth international Yue dialects conference.

Guangzhou, China: Jinan University.

Ohala, J. J. (1986). Against the direct realist view of speech perception. Journal of Phonetics,

14, 75-82.

Ohala, J. J. (1989). Sound change is drawn from a pool of synchronic variation. L. E. Breivik

& E. H. Jahr (Eds.), Language change: Contributions to the study of its causes (pp.

173-198). Berlin, Germany: Mouton de Gruyter.

Smit, A. B., Hand, L., Freilinger, J.-J., Bernthal, J. E., & Bird, A. (1990). The Iowa

Articulation Norms Project and its Nebraska replication. Journal of Speech and

Hearing Disorders, 55, 779-798.

So, L. K.-H. 1993. Cantonese Segmental Phonology Test. Hong Kong: Bradford Publishing

35

Company.

T’sou, B., Lee, T. H.-T., Tung, P., Chan, A., Man, Y., & To, C. K., (2006). Hong Kong

Cantonese Oral Language Assessment Scale. Hong Kong: Language Information

Sciences Research Centre, City University of Hong Kong.

To, C. K. S., Cheung, P. S. P., & McLeod, S. (2013). A population study of children's

acquisition of Hong Kong Cantonese consonants, vowels, and tones. Journal of Speech,

Language, and Hearing Research, 56(1), 103-122.

Wang, L. (1985). Hanyu yuyin shi [The History of Chinese Phonology]. Beijing, China:

Zhongguo Shehui Kexue Chubanshe.

Wong, S.-L. (1941). A Chinese syllabary pronounced according to the Cantonese dialect.

Hong Kong: Chung Hwa Book Co.

Wong, S.-Y. (2005). Phonological variation: the case of the syllable-final velar nasal in

Cantonese. Unpublished master’s thesis. The University of Hong Kong, Hong Kong.

Yeung,. H. S.-W. (1980). Some aspects of phonological variations in the Cantonese spoken in

Hong Kong. Unpublished master’s thesis. The University of Hong Kong, Hong Kong.

Zee, E. (1999a). Chinese (Hong Kong Cantonese). In Handbook of the International Phonetic

Association, 58-60. Cambridge: Cambridge University Press.

Zee, E. (1999b). Chinese (Hong Kong Cantonese). Handbook of the International Phonetic

Association (pp. 58-60). Cambridge: Cambridge University Press.

36

Zhan, B.-H. & Cheung, Y. S. (1987). A survey of dialects in the Pearl River Delta,

comparative morpheme – Syllabary, Vol.1. Hong Kong: New Century Publishing

House.

37

Table 1 Evidence of Nine Phonetic Variations in Hong Kong Cantonese Sound change or phonetic variation

Context Examples Study 1: Diachronic evidence (status) *

Study 2: Synchronic study (status)

1. [n-]→[l-] Syllable initial position

男 ‘boy’ [nаm4]→[lаm4]

Almost complete

Complete Adult: 94.6% Children: 92.6%

2. [ŋ-]→Ø- Syllable initial position

牛 ‘cow’ [ŋɐu4]→[ɐu4]

Incomplete Half-way Adult: 37.5% Children: 71.9%

3. Ø-→[ŋ-] Syllable initial position

屋 ‘house’ [ʊk7]→[ŋʊk7]

Incomplete Half -way Adult: 31.3% Children: 7.1%

4. [kw-]→[k-] ([kwh-]→[kh-])

Before /ɔ/

果 ‘fruit’ [kwɔ2]→[kɔ2]

Incomplete Half -way Adult: 50.9% Children: 42.0%

5. Syllabic [ŋ̍]→Syllabic [m̩] Individual syllable

五 ‘five’ [ŋ̍5]→[m̩5]

Almost complete

Half -way Adult: 95.5% Children: 99.2%

6. [-ŋ]→[-n] Syllable final position

香(蕉) ‘banana’ [hœŋ1]→[hœn1]

Incomplete Half -way Adult: 24.1% Children: 29.7%

7. [-n]→[-ŋ] Syllable 乾 ‘dry’ Incomplete Incomplete

38

final position

[kɔn1]→[kɔŋ1] (rare) Adult: 0.8% Children: 5.0%

8. [-k]→[-t] Syllable final position

腳 ‘foot’ [kœk8]→[kœt8]

Incomplete Incomplete (rare) Adult: 2.6% Children: 2.8%

9. [-t]→[-k] Syllable final position

渴 ‘thirsty’ [hɔt8]→[hɔk8]

Incomplete Incomplete (rare) Adult: 8.0% Children: 14.5%

39

Table 2

Distribution of Age and Gender of the Participants

Age

Range

(yr;mth)

Year born

Mean age N

Total N Male Female Male Female

10;8-12;4 1993-1994 11.26 11.25 70 68 138

18-45 1960-1987 - - 53 59 112

Total 123 127 250

40

Table 3

Mean Percentage in Producing the Non-standard Forms by Male and Female Speakers in the Adult and Child Groups

Elicited syllable

Overall Adult Child

Adult

(n=112)

Child

(n=138) χ2 p-value

Male

(n=53)

Female

(n=59) χ2 p-value

Male

(n=70)

Female

(n=68) χ2 p-value

Phonetic variation 1: [n-]→[l-]

男 [nаm] ‘male’ 94.6% 94.2% 0.03 .87 96.2% 93.2% .08 .78 91.3% 97.1% 2.06 0.15

Phonetic variation 2: [ŋ-]→Ø-

牛 [ŋɐu] ‘cow’ 37.5% 65.0% 18.64 <.01** 45.3% 30.5% 2.01 .16 73.9% 55.9% 4.89 <.03**

Phonetic variation 3: Ø-→[ŋ-]

屋 [ʊk] ‘house’ 31.3% 5.8% 28.12 <.01** 24.5% 37.3% 1.56 .21 4.3% 7.4% .59 .44

Phonetic variation 4: [kwɔ] → [kɔ]

果 [kwɔ] ‘fruit’ 50.9% 37.7% 4.39 .04* 50.8% 50.9% .00 1.00 44.3% 30.9% 2.64 .10

(continued)

41

Elicited syllable Overall Adult Child

Adult Child χ2 p-value Male Female χ2 p-value Male Female χ2 p-value

Phonetic variation 5: syllabic [ŋ̍]→ syllabic [m̩]

五 [ŋ̍] ‘five’ 95.5% 98.6% 2.07 .15 100% 91.5% .92 .09 97.1% 100% 1.97 .16

Phonetic variation 6: [-ŋ] → [-n]

[phʊŋ]a - 0% - - - - - - 0% 0% 0 1.00

蘋 [phɪŋ] ‘apple’ 0 % 0 % 0 1.00 0% 0% - - 0% 0% 0 1.00

糖 [thɔŋ] ‘candy’ 11.6% 13.1% 0.13 .72 15.1% 8.5% .64 .43 7.2% 19.1% 4.23 <.04*

餅 [pɛŋ] ‘biscuit’ 25.9% 33.6% 1.72 .19 30.2% 22.0% .59 .44 30.90% 36.4% 0 1.00

羹 [kɐŋ] ‘spoon’ 33.9% 37.0% .25 .62 35.8% 32.2% .04 .84 31.4% 42.6% 1.86 .17

橙 [tshаŋ] ‘orange’ 35.7% 40.9% .69 .41 32.1% 39.0% .32 .57 27.1% 55.2% 11.17 <.01**

香 [hœŋ] ‘banana’ 37.5% 51.9% 4.88 .03 34.0% 40.7% .26 .61 43.5% 60.6% 3.96 .05**

Average: 24.1% 25.2%

(continued)

42

Elicited syllable Overall Adult Child

Adult Child χ2 p-value Male Female χ2 p-value Male Female χ2 p-value

Phonetic variation 7: [-n]→[-ŋ]

鏈 [lin] ‘zip’ 0% 0.7% 0.83 .36 0% 0% - - 0% 1.5% 1.04 .31

圓 [jyn] ‘circle’ 0% 0.7% 0.82 .37 0% 0% - - 0% 1.5% 1.02 .31

門 [mun] ‘door’ 0% 0% 0 1.00 0% 0% - - 0% 0% 0 1.00

信 [sɵn] ‘letter’ 0.9% 2.2% 0.67 .41 1.9% 0% 0 .96 4.4% 0% 3.07 .08

裙 [kwhɐn] ‘dress’ 0% 1.5% 1.65 .20 0% 0.0% - - 2.9% 0% 2.06 .15

蛋 [tаn] ‘egg’ 0.9% 0.7% 0.02 .89 0% 1.7% 0 1.00 1.5% 0% 1.07 .30

乾 [kɔn] ‘dry’ 7.1% 21.9% 10.37 <.01** 7.5% 6.8% 0 1.00 16.2% 27.5% 2.58 .11

Average: 1.3% 4.0%

(continued)

43

Elicited syllable Overall Adult Child

Adult Child χ2 p-value Male Female χ2 p-value Male Female χ2 p-value

Phonetic variation 8: [-k]→[-t]

色 [sɪk] ‘color’ 0% 0.8% .35 .87 0% 0% - - 0% 1.6% 1.07 .30

屋 [ʊk] ‘house’ 0% 0% 0 1.00 0% 0% - - 0% 0% 0 1.00

腳 [kœk] ‘feet’ 8.0% 6.6% 0.20 .66 7.5% 8.5% 0 1.00 4.3% 8.8% 1.12 .29

落 [lɔk] ‘drop’ 20.9% 14.5% 1.76 .19 17.6% 23.7% .61 .43 10.0% 19.1% 2.31 .13

Average: 7.2% 5.5%

Phonetic variation 9: [-t]→[-k]

熱 [jit] ‘hot’ 0% 0.7% .83 .36 0% 0% - - 0% 1.5% 1.07 .30

滑 [wаt] ‘smooth’ 0% 0.7% .82 .37 0% 0% - - 0% 1.5% 1.04 .31

渴 [hɔt] ‘thirsty’ 24.1% 38.7% 6.0 .01* 28.3% 20.3% .58 .45 39.1% 38.2% 0.01 .91

一 [jɐt] ‘one’ 10.7% 11.6% .05 .83 7.5% 13.6% 1.06 .30 11.4% 11.8% 0.00 .95

Average: 8.7% 12.9%

Note: * p-value <.05. ** p-value <.01. a Results of /phʊŋ/ were collected from the Nonword Repetition Test with Pseudosyllables administered on the same group of children.